Dispersal can strongly influence local population dynamics and may be critical to species persistence in fragmented landscapes. Theory predicts that dispersal by resident stream organisms is necessary to offset the loss of individuals to downstream drift. However, there is a lack of empirical data linking dispersal and drift to local population dynamics in streams, leading to uncertainty regarding the general demographic significance of these processes and the power of drift to explain observed dispersal patterns. I assessed the contribution of dispersal along a first-order stream to population dynamics of the headwater salamander Gyrinophilus porphyriticus (Plethodontidae). I conducted mark–recapture surveys of two contiguous 500 m long sections of a study stream in June, July, and August of 1999, 2000, and 2001. Movement by G. porphyriticus larvae and adults showed a strong upstream bias in the study stream, as well as in 11 other streams that I surveyed. Using mark–recapture models and Akaike's information criterion for model selection, monthly probability of dispersal from the downstream section to the upstream section of the study stream (mean ± 1 se) was estimated to be 0.02 ± 0.01. The probability of dispersal from the upstream section to the downstream section was 0.00 ± 0.00. Monthly survival probabilities did not differ between sections. Estimated monthly population growth rates were 1.01 ± 0.01 in both sections. Net dispersal from the downstream section to the upstream section contributed to the equality of population growth rates. Additionally, reproduction and individual condition were lower in the upstream section, suggesting that population stability and growth there (i.e., λ ≥ 1.0) may have been dependent on immigration from downstream. Similar interactions between dispersal and variation in local demographic rates along the stream continuum may underlie the distribution and abundance patterns of other organisms at multiple spatial scales. Greater empirical understanding of these interactions will improve conservation of stream biota. The strong upstream bias of G. porphyriticus movement suggests that dispersal is not a response to downstream drift. I propose an alternative model for the evolution of directionally biased dispersal based on the landscape-scale distribution of headwater habitats.